SEGMENTATION OF TOUCHING CELLS USING GRADIENT FLOW TRACKING

The development of automated and robust computational algorithms for 3D cell segmentation remains challenging in situations where the cells are touching each other or connected together. In this paper, we present a novel automated method that aims to tackle the aforementioned challenges in the segmentation of clustered or connected 3D cells. We first diffuse the gradient vector field with an elastic deformable transform. Then, a gradient flow tracking procedure is performed for each voxel to find the corresponding center that the point flows to. All points that flow to the same center are grouped as a region. In this way, the boundaries between touching cells are formed naturally, and the touched cells are divided. Evaluation results from 3D image data of zebrafish and C. Elegant indicate good performance of the proposed method